It is the 25th of July, 2525. Now that the usage of Large Language Models (LLMs) is so ubiquitous, it has become nearly impossible to find actual human-made articles on the internet and not get lost in all the gibberish produced by LLMs.
(Un)fortunately, LLM technology has regressed significantly. Neural networks have been trained on data that is mostly generated by older LLMs, which in turn were trained on even older recycled data. As a result, the output produced by most LLMs is a long string of lowercase letters, each chosen uniformly at random and independently of the others.
You decided to make a program to scavenge the internet in search of human-made articles. Your program must determine whether a given text is human-made or generated by an LLM.
A given text is guaranteed to be exactly one of the following:
The input consists of:
Your submission will be run on exactly $$$100$$$ test cases, all of which will have $$$|s| = 3\cdot10^5$$$ and $$$n = 5000$$$. The samples are smaller and for illustration only.
For each test case where $$$s$$$ is human-made, $$$s$$$ is fixed and does not change between each of your submissions. For each test case where $$$s$$$ is generated by an LLM, each of your submissions will receive a new string $$$s$$$, generated from independently and uniformly picking random English lowercase letters (a-z).
If the given string was human-made, output "yes". Otherwise, if it was generated by an LLM, output "no".
ballooncodingballoonacceptedchallengechallengecoding 5 accepted balloon challenge coding algorithms
yes
nlaiueakuyclocedxlwvxdbiifqjbinucjabzyaqakhohgympi 5 aaaaaaaaaa bbbbbbbbbb aababbb aaabbbaaa abaaba
no
aaaaaaaaaaabbbbbbbbbbaaaaaaaaaaaaaaabbbbbbbbbbbbbb 5 aaaaaaaaa bbbbbbbbbb aabbbbb aaaaabbbb bbbbba
yes
| Name |
|---|


